基于神经网络的企业内部网络恶意入侵检测方法  

Enterprise Internal Network Malicious Intrusion Detection Method Based on Neural Networks

作　　者：彭国勇 PENG Guoyong(Jiangxi Tobacco Company Yichun City Company,Yichun Jiangxi 336000,China)

机构地区：[1]江西省烟草公司宜春市公司,江西宜春336000

出　　处：《信息与电脑》2023年第19期65-67,共3页Information & Computer

摘　　要：针对现有检测方法准确率低的问题,提出基于神经网络的企业内部网络恶意入侵检测方法。首先,对企业内部网络数据进行归一化和标准化处理,消除误差,生成独立分布的对抗样本。其次,把数据流输入神经网络中,按照数据流的顺序进行排序,通过比较差异获得恶意入侵特征。最后,进行实验分析。实验结果表明,该方法的检测准确率高于对比方法,能够准确识别数据集中的攻击行为。Aiming at the problem of low accuracy in existing detection methods,a neural network-based malicious intrusion detection method for enterprise internal networks is proposed.Firstly,normalize and standardize the internal network data of the enterprise,eliminate errors,and generate independently distributed adversarial samples.Secondly,input the data stream into the neural network,sort it in the order of the data stream,and obtain malicious intrusion features by comparing differences.Finally,conduct experimental analysis.The experimental results show that the detection accuracy of this method is higher than that of the comparison method,and it can accurately identify attack behaviors in the dataset,which is better than the control group.

关 键 词：神经网络 内部网络 恶意入侵 检测 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]